Kalle and Lennert reported problems with the new work
code, and at least Kalle's problem I was able to trace
to a missing jiffies initialisation.
I also ran into a problem where occasionally I couldn't
connect, which seems fixed with kicking the work items
after scanning.
Finally, also add some sanity checking code to verify
that we're not adding work items while an interface is
down -- that case could lead to something similar to
what Lennert was seeing.
There still seems to be a race condition that we're
trying to figure out separately.
Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Tested-by: Lennert Buytenhek <buytenh@marvell.com>
Tested-by: Kalle Valo <kalle.valo@iki.fi>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
        ieee80211_mlme_notify_scan_completed(local);
        ieee80211_ibss_notify_scan_completed(local);
        ieee80211_mesh_notify_scan_completed(local);
+       ieee80211_queue_work(&local->hw, &local->work_work);
 }
 EXPORT_SYMBOL(ieee80211_scan_completed);
 
 
                    wk->chan == local->tmp_channel &&
                    wk->chan_type == local->tmp_channel_type) {
                        wk->started = true;
+                       wk->timeout = jiffies;
                }
 
                if (!wk->started && !local->tmp_channel) {
        if (WARN_ON(!wk->done))
                return;
 
+       if (WARN_ON(!ieee80211_sdata_running(wk->sdata)))
+               return;
+
        wk->started = false;
 
        local = wk->sdata->local;